Ramp-down
0 200 400 600 800
-0.8 -0.7 -0.6
-0.5
SiGe + implant + RTA
A.U. of Stress Rel a ted
Temperature (ºC) Ramp-up
SiGe Si SiGe
Si SiGe
Si
Defect-free in the Si substrate
36
To investigate the relationship between the formation of defects in the underlying Si substrate and the combination of implanted SiGe and MSA process on the pattern wafer, identical processing conditions were applied to blanket Si (001) wafers and the corresponding optically measured wafer bow height, which revealed the degree of wafer warpage [14-15], was as plotted in Fig. 2.4.6 The strong correlation between wafer bow height and density of defects observed in the Si substrate was well established. From Fig. 2.4.6, the 100nm-thick strained-Si1-xGex (x nearly 0.35) introduced a compressive film stress with wafer concave downward, then the subsequent medium implantation projection range (Rp) using Arsenic (As), at an energy of approximately 50keV and a dose of 3E13cm-2 to make species to reach medium position of the strained-SiGe layer, damaged the upper portion of the strained-SiGe layer, and partially relaxed stress in the strained-SiGe, reducing wafer warpage. Following spike RTA, the implant-damaged strained-SiGe layer was repaired, but not completely. The subsequent MSA caused significant wafer warpage and plastic deformation from its initial compressive to its final high tensile state. Then, defects in the underlying Si substrate were observed as shown in the TEM image. Meanwhile, large wafer warpage up to 150m in tensile state impacted critical lithography stage, such as contact-hole lithography. For comparison, a strained-SiGe sample directly underwent the MSA process, exhibiting no change in bow height, while its underlying Si was defect-free and did not have warpage-induced lithographic limitation. This result implies that some relaxation of strain
37
SiGe caused by implantation considerably changes the bow height and causes the formation of defects in the underlying Si for strained-SiGe that undergoes MSA as shown in Fig2.4.7.
-50 0 50 100 150
SiGe Implant RTA MSA
Blanket wafer,BowHeight (um) SiGe + Implant + RTA + MSA
Thickness=100nm, Ge 35 at.%
Tensile
SiGe Implant RTA MSA
Blanket wafer,BowHeight (um) SiGe + Implant + RTA + MSA
Thickness=100nm, Ge 35 at.%
Tensile
Figure 2.4.6 Change in bow height of strained-SiGe wafer associated with following implantation, RTA and subsequent MSA. A fully strained-SiGe sample directly underwent MSA, exhibited no change in bow height.
Figure 2.4.7 The combination of implanted strained SiGe and MSA form defects in the underlying Si substrate
38
2.4.2. Dopant species effect on strain relaxation of SiGe and device performance
To eliminate the relaxation of strain SiGe by post-SiGe medium implantation range Rp of n-type species, As, implantation conditions are modified to prevent strain relaxation, defect formation and warpage-induced lithographic limitation. Accordingly, a lighter atom, phosphorus (P), with energy of around 20KeV with tilted 25ﾟand a dose of 4e13cm-2 was used to give the same medium implantation range Rp of As, and minimize the SCE properties. In the inset of Fig. 2.4.8, TEM image indicates that a medium implantation Rp of the species, As, was employed into strained-SiGe layer and led to a clear pre-amorphous implant (PAI) layer in the SiGe layer. Varying As implant dosage in the SiGe layer to cause different level of PAI was investigated as shown in Fig.2.4.9. It was concluded that As implant dosage over 3E13cm-2 would cause PAI layer in the strained SiGe. The result was also feed to simulation result by the kinetic Monte Carlo (kMC) model. The degree of the As induced implantation damage in SiGe was defined during simulation that Si atoms had moved one-third of the way from original position toward vicinity [16-18]. Since As atoms are large, the damaged layer of pre-amorphorized implantation resulted in the large relaxation of strain SiGe and severely affected the pseudomorphic SiGe. The following RTA process was unable to recover relaxed strained-SiGe, which was indicated from the broadening XRD rocking curve of SiGe in Fig. 2.4.10. Therefore, subsequent MSA produces defects in the underlying Si substrate, as revealed by the XRD rocking curve of the broadening Si peak in
39
0.00 0.02 0.04 0.06 0.08
1E16
P induced damage P concentration
0.00 0.02 0.04 0.06 0.08
1E16
P induced damage P concentration
SiGe Si
Si
PAI layer SiGe
0.00 0.02 0.04 0.06 0.08
1E16
P induced damage P concentration
0.00 0.02 0.04 0.06 0.08
1E16
P induced damage P concentration
Full Amorphous layer
Full Amorphous layer
As-grown SiGe As IMP (1e13cm-2)
HR-TEM
As IMP (3e13cm-2) As IMP (6e13cm-2)
Si Si
Full Amorphous layer
Full Amorphous layer
As-grown SiGe As IMP (1e13cm-2)
HR-TEM
As IMP (3e13cm-2) As IMP (6e13cm-2)
Figure 2.4.8 Simulated that implant species arsenic (As) and phosphorus (P) at the comparable implant energy and concentration caused different degrees of damage in the strained SiGe layer
Figure 2.4.9 The As implant dosage effect on PAI layer formation in SiGe
40
Fig. 5(b). Nevertheless, phosphorus, which has a small atomic size as compared to As, implanted at a concentration close to that of the As atoms, did not cause PAI in the strained-SiGe layer, as shown in Fig. 2.4.8. Therefore, the subsequent spike RTA process enabled the non-PAI SiGe to be almost fully recovered as strained-SiGe, and the following MSA did not cause defects in the underlying Si, as revealed by the tight SiGe profile and shaper Si peak in the XRD rocking curve of phosphorus. Figure 2.4.10 also presents results of a simulation of fully strained-SiGe samples as a reference. The figure reveals that the strained-SiGe was almost fully recovered. Hence, well-controlled implantation in the source/drain SiGe is required to prevent strain relaxation and defect formation.
Properly selecting implanted species to form almost fully strained-SiGe enables Si channel stress to be increased to further boost the device drive current. In Fig. 2.4.11, the Raman measurements reveal the relaxation of strained-SiGe by various implanted species on the Si channel stress of the Si substrate. Post-SiGe arsenic implantation caused large strained-SiGe relaxation, reducing the effective stress in the Si channel below that associated with SiGe relaxation by phosphorus.
41
SiGe SiGe SiGe SiGe Si
STI STI STI
Si
Shifted Wave Number (cm-1)
Distance of Raman measurement on pattern wafer (um)
SiGe + As + RTA + MSA
SiGe SiGe SiGe SiGe Si
STI STI STI
Si
Shifted Wave Number (cm-1)
Distance of Raman measurement on pattern wafer (um)
SiGe + As + RTA + MSA SiGe + P + RTA + MSA
scan
Figure 2.4.10 The relaxation of strained-SiGe is a function of dopant impurities, determined from the XRD rocking curve.
Figure 2.4.11 Raman measurements (wave number shifts) indicated the relaxation of strained-SiGe is associated with post implanted species, As and P, and their effects on the Si channel stress in Si
42
The 10% higher drive current (Id,sat) gain of PFET devices versus Lmin (minimum gate length) associated with post-SiGe phosphorus due to less strain relaxation of S/D SiGe than that associated with arsenic in Fig. 2.4.12, was consistent with Raman measurements. Fig. 2.4.13 plots the relationship of the dopant species and the junction leakage in PMOSFET devices. Upon RTA and MSA thermal treatment, post-SiGe phosphorus implantation can result in an almost fully strained-SiGe layer and does not form defects in the underlying Si substrate. Minimized the leakage current to a level comparable to those of PMOSFET devices are fabricated using a non-SiGe process.
10%
Drive current, Ion (mA/um)
Drive current, Ion (mA/um)
Figure 2.4.12 Drive current, Ion-Lmin was affected by the implanted As and P induced damage effect in the strained-SiGe layer, which causes various degrees of strain relaxation upon the RTA and MSA.